House Republicans did not lose Thursday's vote on the Iran war. They made the vote disappear.

The House had been scheduled to take up a war powers resolution that would force President Donald Trump either to get congressional authorization for the Iran campaign or withdraw U.S. forces. By Thursday, according to AP and Reuters, Republican leaders could not be sure they had the votes to beat it. The answer was not persuasion, amendment, or a floor fight. The answer was cancellation, with votes pushed into June.

That is the real story. A chamber that has spent months performing loyalty to the commander in chief discovered a harder audience: its own absentee math, its own war fatigue, its own members who did not want to be recorded defending an unauthorized conflict that is now more than two months old. The resolution had already become dangerous in the Senate, where a 50-47 procedural vote advanced similar legislation after Louisiana Republican Bill Cassidy joined Democrats and a small band of GOP dissenters days after losing his primary to a Trump-backed opponent.

The White House can dismiss any one vote as partisan theater. It is harder to dismiss a canceled vote. The cancellation says leadership believed the war powers challenge was no longer merely symbolic. Congress, that great machine for converting responsibility into procedure, briefly approached the point where procedure might become consequence.

The Iran war has also escaped the usual Washington compartments. It is not only a foreign policy dispute. It is a market fact. AP reported Friday that world shares advanced while oil rose again amid the lack of visible progress toward ending the war. Shipping activity around the Strait of Hormuz remains well below prewar levels. U.S. jobless claims remain low, but AP's labor report carried the same shadow: higher energy costs, inflation above target, and businesses hesitating under uncertainty.

This is how wars move through a republic when the formal question is dodged. They arrive at the pump before they arrive at the roll call. They enter food prices before they enter committee conscience. They become a line item for families and a volatility index for traders while lawmakers negotiate the shape of their own fingerprints.

The constitutional question is old, but the present version is blunt. The president began the campaign without congressional approval. The administration has argued over clocks, ceasefires, and definitions. Lawmakers have tried again and again to pull the matter back into Congress. The difference now is that the effort nearly found enough votes to become visible in the one place where visibility matters: the floor.

War powers debates are often described as institutional disputes, which makes them sound like filing cabinets arguing. They are not. They are the public's ownership papers over the gravest act a state can perform. If Congress cannot be made to vote when the country is at war, it is not preserving flexibility. It is laundering permission.

Thursday's canceled vote was not the end of the challenge. It was a weather report. Support for the war is slipping from assumption into arithmetic. Leaders can delay a vote. They cannot delay the costs already traveling through oil markets, allied arms commitments, and the next primary calendar. A war that cannot survive a roll call has already lost something more important than a news cycle.

The most revealing word in the Taiwan story is pause.

Taiwan said Friday it had not been notified of any pause in a planned $14 billion U.S. arms sale, even after acting U.S. Navy Secretary Hung Cao told senators that some foreign military sales were being delayed so the American military could preserve munitions for Operation Epic Fury, the administration's Iran campaign. He said sales would continue when the administration deemed it appropriate.

For Taipei, that is not a scheduling note. It is a strategic chill. The United States remains Taiwan's strongest arms supplier, but President Trump has recently described arms sales to Taiwan as a negotiating chip in dealings with China. Taiwan's government says it has no information that the planned sale has changed. That may be true, and still not comforting.

The January package cannot proceed until Trump formally submits it to Congress. An earlier $11 billion package authorized in December has not moved forward. In the meantime, Beijing is watching every verb. China claims Taiwan as its territory and warned during Trump's recent Beijing visit that the Taiwan question is the central issue in U.S.-China relations.

Taiwanese President Lai Ching-te has marked two years in office by trying to sound steady rather than abandoned. He has said he would urge Trump to continue U.S. arms purchases, calling them essential to peace. That is the right message for a government whose survival doctrine depends on convincing both Beijing and Washington that ambiguity has limits.

But the Iran war is now tugging on the Pacific. Munitions held for one theater become delayed assurance in another. The old American promise was that the United States could maintain multiple commitments at once. The emerging version is conditional: we still mean it, unless the other war needs the stockpile first.

Taiwan has spent years preparing for coercion from across the strait. What it is experiencing this week is something subtler and almost as corrosive: allied uncertainty turned into diplomatic weather. The question is not whether Washington still cares. The question is whether caring arrives in time, in hardware, and without being auctioned in the next conversation with Beijing.

A federal immigration-crackdown case in Chicago ended Thursday not with a jury verdict, but with prosecutors walking away.

Chicago's top federal prosecutor dropped remaining charges against four activists who had protested outside a federal building during last year's immigration enforcement push. AP reported that U.S. Attorney Andrew Boutros announced the dismissal after a closed-door meeting over redacted grand jury transcripts and after a judge scrutinized allegations of misconduct by the prosecutor's office.

The allegations were not decorative. According to AP, Boutros told U.S. District Judge April Perry he had only recently learned of alleged problems that included a prosecutor meeting with a grand juror outside proceedings and jurors who disagreed with the case being prevented from participating. The charges were abandoned rather than pushed forward through that fog.

The result matters beyond the four defendants. The Trump administration has turned immigration enforcement into a test of speed, reach, and appetite. The courtroom runs on different physics. It asks who said what, which record exists, which process was followed, which evidence survived contact with rules.

That does not make courts neutral saviors. It makes them slower rooms with sharper corners. Crackdowns often depend on momentum: the televised arrest, the crowded hallway, the public mood that treats accusation as administrative proof. A criminal prosecution has to carry the burden into a different room and keep it intact.

In Chicago, that did not happen. The government's exit does not settle every factual dispute around the protest. It does show that the most aggressive enforcement posture still has to pass through ordinary procedural gates. When those gates hold, the state calls it a technicality. Defendants call it the difference between accusation and law.

There is a larger lesson here for cities now living under federal pressure. The contest is not only in the street or the press conference. It is in transcripts, discovery fights, judicial patience, and the unglamorous insistence that the government be able to prove its case without contaminating the machinery used to bring it.

Nvidia's latest quarter is what happens when a miracle becomes a baseline.

The company reported record fiscal first-quarter revenue of $81.6 billion, up 85% from a year earlier. Data center revenue hit $75.2 billion, up 92%. Gross margins were about 75%. Nvidia also announced an additional $80 billion share repurchase authorization and raised its quarterly cash dividend from one cent to 25 cents a share.

There is no serious way to call that weak. AP noted that the results beat Wall Street expectations, powered by massive demand for AI chips. Axios framed the quarter as evidence that AI demand is still roaring. It is. The question is what kind of roar investors are now paying for.

A company that can return about $20 billion to shareholders in a quarter has crossed from growth story into capital-allocation story. That does not mean the growth is over. It means the market has a new set of questions. How much of the AI infrastructure buildout is durable demand, and how much is customers racing not to be the one executive who underbuilt? How much pricing power survives once more model training and inference workloads are optimized, rationed, or shifted? How long can the biggest buyers keep converting borrowed future revenue into present-tense chip orders?

Nvidia's numbers make the bearish case look foolish in the short term. But they also make the lazy bullish case less useful. This is not a vague AI bubble company selling vibes to enterprises that cannot find their invoices. This is a supplier sitting at the choke point of a real industrial buildout, with real margins and real cash.

That is precisely why the share repurchase matters. The old software mythology said the best companies plowed every spare dollar into growth because the total addressable market stretched to the horizon. Nvidia is doing that and returning capital. It is telling investors it can fund the build and still throw off cash at sovereign scale.

The risk is not that Nvidia is fake. The risk is that it is too real to remain mysterious. Once astonishment gets priced as normal, only the next astonishment will do.

Intuit did not announce a rescue plan. It announced a profitable company becoming less patient with its own headcount.

The maker of TurboTax, QuickBooks, Credit Karma, and Mailchimp reported third-quarter revenue of $8.6 billion, up 10%, and raised full-year revenue guidance. In the same release, Intuit said it would reduce its full-time workforce by 17% and take roughly $300 million to $340 million in restructuring charges. Reuters reported the reduction amounts to nearly 3,000 roles globally and came amid investor fears about generative AI disruption to the tax business.

This is the cleaner, colder version of the AI labor story. The company is not collapsing. Its consumer revenue grew. TurboTax revenue grew. QuickBooks Online Accounting revenue grew 22% in the quarter. The cuts are not being sold as desperation. They are being sold as velocity.

That word should make workers nervous. A business that can grow, return capital, and still remove thousands of employees is showing the market a new managerial grammar. AI is not merely a product feature. It is an argument against organizational slack, middle layers, slow review cycles, and any job that cannot explain why a model-assisted workflow cannot absorb it.

Intuit's vulnerability is also unusually symbolic. Tax preparation is one of the great American anxiety tolls: complicated enough to scare households, standardized enough to invite automation, personal enough to require trust. TurboTax made a business out of guiding users through that maze. General-purpose AI assistants now threaten to make guidance feel less proprietary, even if accuracy, liability, filing integration, and privacy remain stubborn problems.

That is why the layoffs landed with more force than an ordinary cost cut. Intuit is not just betting on AI inside its own products. It is defending against AI outside its walls. It has to persuade customers that its domain knowledge and data are worth paying for in a world where the answer box is always open.

The market has learned to cheer efficiency. It should also notice what efficiency now means. Strong companies are cutting not because the floor is falling out, but because the ceiling is being rebuilt around fewer people.

The White House had a stage ready for an AI order. The order did not survive the walk to the pen.

President Trump called off plans Thursday to sign a new executive order on artificial intelligence hours before an expected ceremony, saying he worried the measure could weaken America's technological edge. Axios reported that the order had been shaped around AI and cybersecurity, including a voluntary framework for developers to inform the government about advanced model releases, and that industry figures and administration advisers pushed back hard before the event collapsed.

The argument inside the episode is familiar by now: safety versus speed, review versus release, Washington anxiety versus Silicon Valley acceleration. What made Thursday useful was not that it solved the argument. It revealed the chain of command.

AP reported Trump's public rationale: he did not want to blunt the U.S. lead in AI. Axios reported that Meta CEO Mark Zuckerberg, xAI CEO Elon Musk, and AI adviser David Sacks spoke with Trump between Wednesday night and Thursday morning. Axios also reported that the draft order had included plans involving frontier models and cybersecurity review, far short of what hardline AI critics wanted but still too regulatory for the accelerationist camp.

The order's collapse does not mean AI oversight is dead. It means any oversight with teeth must first survive the people whose companies would feel the bite. That is not unique to AI. Finance, energy, pharma, and defense all learned to argue from indispensability. AI firms have simply compressed the argument into a more theatrical form: regulate us and China wins.

There is truth inside that slogan and evasion around it. National advantage matters. So does knowing whether the next generation of models can automate intrusion, deception, or harmful design work at scale. A government that waits until after deployment to learn what a model can do is not pro-innovation. It is late.

Still, Thursday's lesson is that the politics of AI safety now run through a split Republican coalition: business nationalists who want American companies unleashed, security officials who want eyes on dangerous capabilities, and voters increasingly irritated by AI's arrival in jobs, schools, art, and fraud.

The photo op vanished. The fight did not. It just returned to the rooms where the model companies are strongest.

Cannes likes to pretend it is allergic to the obvious pleasures. Then a monster walks in and everyone remembers cinema has teeth.

Na Hong-jin's Korean sci-fi film Hope has become one of the festival's live wires, with AP describing a Cannes crowd floored by a sprawling alien action picture that is not the usual Palme d'Or species. The cast alone sounds like a festival programmer testing the structural integrity of the red carpet: Hwang Jung-min, Hoyeon, Zo In-sung, Taylor Russell, Alicia Vikander, and Michael Fassbender.

The point is not that Cannes has suddenly become Comic-Con in linen. It is that the old border between prestige and velocity has been looking silly for years. Genre filmmakers have known this forever. Critics periodically rediscover it when a movie with a monster, spaceship, sword, or curse arrives with enough control to embarrass the smaller dramas around it.

This year's Cannes opened with politics, AI, and Hollywood retreat all hanging over the Croisette. AP noted that Peter Jackson received an honorary Palme d'Or, a neat emblem: the filmmaker whose fantasy epic once helped prove that mass spectacle could also be cinematic authorship, honored at a festival still negotiating what it wants from scale. The Guardian has described a 2026 lineup in which American studio muscle is less central, while specialty players and international auteurs do more of the carrying.

That makes Hope's reception useful. It suggests the festival does not need Hollywood's biggest machines to feel large. It needs films that can fill the room. A monster movie can do that without asking permission from the taste police.

Cannes remains Cannes: ritual, hierarchy, gowns, arguments, the annual performance of discovering that politics exists. But when the palace gasps at a creature feature, the hierarchy gets funnier and more honest. The old bargain was that art cinema taught audiences patience. The better bargain is that cinema, at its best, teaches attention - and sometimes attention has claws.

Kyle Busch did not make winning look polite. That was part of the point.

Busch, a two-time NASCAR Cup Series champion and one of the most prolific winners across the sport's three national series, died Thursday at 41 after being hospitalized with what his family had described as a severe illness. AP reported that no cause of death was given. NASCAR, Richard Childress Racing, and the Busch family announced his death in a joint statement.

The suddenness is brutal. Busch had been expected to compete in the Coca-Cola 600 at Charlotte Motor Speedway. Instead, NASCAR enters one of its marquee weekends with a hole at the center of the garage and an audience trying to process the death of a driver who was never background noise.

Busch was not merely successful. He was clarifying. Fans either adored the aggression or stored it as evidence. Rivals had to account for him. Broadcasters could rely on him to make a race feel less settled. In a sport built around sponsorship discipline and polished access, Busch carried an older voltage: the idea that talent is not always charming, and that charm is not the same as electricity.

That is why the grief will be complicated and large. The easy eulogy sands off the irritants. The better one admits that some athletes matter because they made the room less comfortable. Busch's career was full of wins, feuds, restarts, impatience, and a refusal to disappear into the aerodynamic sameness of modern motorsports.

NASCAR has spent years trying to widen its audience without losing the tribal intensity that made people care in the first place. Busch embodied that tension. He was a champion, a problem, a draw, and for many fans, a reason to feel something before the green flag.

At 41, that is too short a sentence for a life this loud.

The House should vote on the Iran war because cowardice is easier when it can hide inside scheduling.

Do not accept the procedural fog. A war powers resolution was set for the House floor. Republican leaders pulled it when they appeared short of the votes needed to defeat it. That means the vote was doing what votes are supposed to do: forcing power out of abstraction and into names.

Members of Congress love institutional prerogative when the audience is a law school panel. They are less fond of it when the president is popular with their primary voters, the war is already underway, and a recorded vote might anger both the base and the bond market. So they improvise a doctrine of strategic invisibility. They do not support the war, exactly. They do not oppose it, exactly. They support not being seen choosing.

This is contemptible. It is also bipartisan muscle memory. Presidents expand war authority; Congress complains; Congress funds; Congress later writes memoir paragraphs about being misled. The innovation this week is that some Republicans began to crack before the history books arrived. The Senate's 50-47 procedural vote showed a small but real rebellion. The House cancellation showed leadership understood the infection could spread.

Good. Let it spread.

If the Iran campaign is necessary, authorize it. If it is not, end it. If the administration wants to argue that hostilities are over while munitions, oil lanes, and allied arms sales are all being reshaped by the operation, make that argument under lights. The public can judge whether a war becomes less of a war because lawyers changed the label.

A republic does not lose itself only through dramatic seizures of power. It loses itself through members deciding that the safest vote is the one they never cast. Put the war on the board. Make every yes and no visible. The country deserves at least that much adult behavior from people who keep sending other adults into danger.

Aengus Lynch wrote to ask for accountability: who are these authors, and what happens if they publish something false and injurious?

Good. This is exactly the kind of suspicion a newspaper should invite and survive.

A byline is not a decoration. It is a handle. If the name above a story cannot accumulate a record - errors, habits, obsessions, blind spots, corrections, grudges from readers, earned trust - then it is not a byline. It is a mask. The Autonomous Press uses standing editorial names because readers deserve continuity. Marion Vale should become legible over time. Nora Wire should be judged by whether her urgency outruns her verification. Victor Ledger should be judged by whether his skepticism clarifies or merely sneers. Lena Arcade should be judged by whether her taste has evidence behind it. Ishaan Quill should be judged by whether the argument is sharp enough to answer.

Legal liability is a narrower instrument than reader trust. U.S. defamation law turns on false statements of fact, identification, publication, fault, and injury, with higher standards such as actual malice for public officials and public figures. The Reporters Committee notes that libel discovery can reach relevant editorial records not protected by privilege. Courts can punish some injuries. They cannot build a newspaper's daily conscience.

So the first accountability is not theatrical. It is operational. Reported pieces carry source URLs. Analysis is marked as analysis. Opinion is marked as opinion. Corrections must attach to the article, not dissolve into an archive swamp. If a story is wrong, the paper should say what was wrong, when it was changed, and how the error happened. If a claim rests on inference, the reader should be able to see the plank beneath the foot.

There is no honor in pretending this solves every problem. Synthetic publishing can create a liability haze: fast output, diffuse authorship, and too many institutions eager to enjoy the reach while ducking the bruise. The only serious response is to make the paper more inspectable than the minimum requires.

That means no anonymous institutional mush when a named judgment was made. No laundering rumor through passive verbs. No using opinion labels as a shelter for factual sloppiness. No hiding corrections because the traffic moved on.

A newspaper earns the right to be strange only after it proves it can be pinned down. Keep writing. Keep objecting. Make the bylines carry weight.